Trippple Checks

[ATH]

Originally Posted by Madpoo

This... this is why independent checks are critically important:
M42525269

I'm doing my own run and I'm pretty sure I'll come up with a different answer.

Can you please remove the 2 successes from Xolotl:
https://www.mersenne.org/report_top_500_ll/

[kladner]

Originally Posted by Madpoo http://www.mersenneforum.org/images/...s/viewpost.gif
This... this is why independent checks are critically important:
M42525269

I'm doing my own run and I'm pretty sure I'll come up with a different answer.

That is pretty glaring. Excellent example.

[Madpoo]

Originally Posted by ATH

Can you please remove the 2 successes from Xolotl:
https://www.mersenne.org/report_top_500_ll/

As soon as my run finishes I'll try to cleanup that mess. I wanted to see if I match the first run (very probable) or if a 5th check would be necessary. My run will finish in another 7 hours (but I probably won't check on it for 12 hours since I imagine I'll be asleep at the time)

For now I'm sure the appearance of 2 successes in that report generated some stir... LOL I didn't even think about the fact that although one result said it was composite (but residue zero), one result said it was prime. Normally a verified prime would still be obscured but apparently this is a funky edge case since the result was "verified" but one of them didn't think it was prime. Does not compute... illogical...

[retina]

Originally Posted by Madpoo

This... this is why independent checks are critically important:
M42525269

I'm doing my own run and I'm pretty sure I'll come up with a different answer.

This would be hilarious if someone found a factor while the "proven" prime results were still there. Would the server have an existential crisis over having both a verified prime and a known factor for the same exponent?

[GP2]

Originally Posted by Mark Rose

i have a machine that seems to be turning in an inordinate of mismatches recently. Could anyone please run triple checks on these exponents?

Your result for M69531503 was a mismatch too.

[Madpoo]

Originally Posted by retina

This would be hilarious if someone found a factor while the "proven" prime results were still there. Would the server have an existential crisis over having both a verified prime and a known factor for the same exponent?

Good question...

Anyway, something that will surprise nobody, my result showed it was not prime and did in fact match that first result:
M42525269

I've fixed those particular results and now I just need to make sure that any reports showing "success" get updated to reflect reality (I think they'll self correct when they update at the next hourly interval).

[Mark Rose]

Originally Posted by GP2

Your result for M69531503 was a mismatch too.

I'm going to have to dig more into that machine next week.

[ATH]

Originally Posted by Madpoo

I've fixed those particular results and now I just need to make sure that any reports showing "success" get updated to reflect reality (I think they'll self correct when they update at the next hourly interval).

That did not work unfortunately.

[Madpoo]

Originally Posted by ATH

That did not work unfortunately.

Well, he's gone and done it again anyway:
M45111893

Sheesh. We were in contact with him a while back to try and figure out what the issue was with the weird machines, but nothing was ever resolved. Might need to reach out again and maybe just convince him to not bother running mprime on those particular systems.

[Prime95]

Originally Posted by Madpoo

This... this is why independent checks are critically important:
M42525269

I'm doing my own run and I'm pretty sure I'll come up with a different answer.

Going forward, the server will now reject reports of a composite with residue '0000000000000000'. It previously only rejected residues of '0000000000000002'.

[retina]

Originally Posted by Prime95

Going forward, the server will now reject reports of a composite with residue '0000000000000000'. It previously only rejected residues of '0000000000000002'.

Nitpicking here, but of course both of those values can be legitimate last-64-bits residues, each with a probability of 1/2^64. I wouldn't bet my last dollar on it but it could happen.